Journal of the Neurological Sciences,
1990, 98:91-97
91
Elsevier JNS 03357
Quantitative analysis of gait in Parkinson patients" increased variability of stride length O. Blin 1,2, A . M . Ferrandez 2 and G. Serratrice ~ 1Clinique des maladies du syst~me nerveux et de l'appareil locomoteur CHU Timone, Marseille (France), and 2URA CNRS 1166, Cognition et mouvement, tray. Susini, Marseille (France)
(Received 4 December, 1989) (Revised, received 19 March, 1990) (Accepted 20 March, 1990)
SUMMARY Analysis of the spatio-temporal and kinematic parameters of locomotion recorded in 21 parkinsonian patients compared to 58 normal elderly subjects showed significant differences in all the recorded parameters. However the relationship between these parameters was preserved, as was the basic locomotor pattern. The variability of stride length, more marked in parkinsonian patients, increased as a function of the clinical stages of Hoehn and Yahr. This index could be useful in assessing the course of the disease in patients.
Key words: Parkinson's disease; Locomotion; Quantitation
INTRODUCTION In spite of the frequency of gait disorders in Parkinson patients, few quantified studies of locomotion have been carded out. Differences between the locomotor parameters in parkinsonian patients and control subjects of the same age have already been reported (Knutsson 1972; Murray et al. 1978; Stem et al. 1983) for walking velocity, stride length and duration, stance duration and double support duration. Normally, there are strong correlations between velocity and different locomotor Correspondence to: Dr. O. Blin,Cliniquedes maladiesdu syst~menerveuxet de l'appareillocomoteur, CHU Timone, F-13385 MarseilleCedex5, France.
0022-510X/90/$03.50 © 1990Elsevier Science Publishers B.V. (BiomedicalDivision)
92 parameters such as stride length, relative double support duration (Ferrandez et al. 1988). These links may be called "invariant patterns". In fact, Stem et al. (1983) showed that "in parkinsonian patients the relative contributions of pace-length and cadence to the velocity of walking are no different from those in the elderly population at large". These authors even suggested quantifying only walking velocity to evaluate the locomotor disability of parkinsonians. Thus, certain invariants are preserved in parkinsonian patients. Locomotor parameters are different, but the relationship between the velocity and these parameters is normal. Besides, in these studies, no specific signs of Parkinson's disease have been detected in walking. Similar disorders have been observed in patients with very different pathologies (Imms and Edholm 1981). At the within-trial level, increased variability of movement exhibited by parkinsonian patients may be related to an increased variability in force production (Sheridan et al. 1987). Furthermore, Parkinson patients exhibit a force production impairment and have more irregular force-time curves than controls (Stelmach et al. 1989). The variability of some locomotor parameters was measured by Gabell and Nayak (1984) for a highly selected elderly population (only 32 out of 1187 subjects). They found no differences in variability for any of the parameters measured and concluded that any increases in variability of gait parameters must be due to pathology. We computed these variabilities on stride length and stride duration Parkinson patients. The relationship between progressive Parkinson's disease and the patient's walking difficulty has not been sufficiently investigated. Murray et al. (1978) showed that most locomotor parameters are related to the stages of the disease described in his clinical classification, although they admit that this classification is not very accurate. Stern et al. (1983) showed that a relationship exists between the different stages of the Hoehn and Yahr classification and factors such as walking velocity and duration of half turn. The aim of this study was: (1)to compare locomotor parameters and "invariant patterns" in parkinsonian patients and normal elderly controls; (2)to assess the intraindividual variability in parkinsonian patients and controls; (3)to compare parkinsonian patients to each other as a function of their clinical stage, and if possible also to determine evolutive and prognostic features.
MATERIALS AND METHODS
The study deals with 21 parkinsonian patients: 12 females and 9 males between the age of 50 and 85 years (average age 69.6 years), 1.50-1.85 m (average height 1.69 m), weighing between 37 and 88 kg (average weight 65 kg). The diagnosis of idiopathic Parkinson's disease was confn'med by a general and neurological examination. There were no neuromuscular, osteoarticular or psychiatric disorders in any of the subjects. The history of the disease ranged from 1 to 17 years and patients had been given L-DOPA treatment since the onset. According to the Hoehn and Yahr classification (1967) there were 4 patients in stage I, 3 patients in stage II, 7 patients in stage III, and 7 patients in stage IV. There were "freezing" episodes in 5 patients, dystonia in 3, dyskinesia in 5 and " O n - O f f ' episodes in 2.
93 Fifty-eight normal subjects, 23 males and 35 females between the age of 60 and 92 (average age 72 years), 1.52-1.82 m (average height 1.63 m), weighing between 40 and 120 kg without any neurological or rheumatologic disease or treatment liable to affect walking and balance, served as controls.
Procedure and recording The locomotor parameters were automatically recorded by means of an apparatus designed by Bessou et al. (1989). This device measures the longitudinal displacement of both feet during locomotion, and functions as follows. Each foot is attached to a separate string which is pulled as the subject walks. The string unwinds to the required length by means of a gearing-down pulley system. Movement is recorded in the form of an electric signal via a pulley linked to a potentiometer. The characteristics of locomotor displacement (stride length, cycle, stance, swing and double support durations, stride and swing velocities) are measured over a long distance (more than 10 m). This device does not require any particular walkway or specific illumination, and can be used to record normal locomotion. In fact the value of this device has been mainly demonstrated in pathological subjects by Bessou et al. (1989). In our study, date were recorded at 75 Hz, then filtered using a FIR filter with a 33-point window and a cutoff at 10 Hz. The calibration of the device defines the volt/meter coefficient to be used in computing spatial and velocity data. From these values, spatio-temporal data (stride length and stride duration, stance duration and double support duration) and kinematic data on locomotion were calculated (sequence velocity, swing velocity, peak velocity value). The variability index (as defined by Gabell and Nayak 1984) was used to assess intra-individual spread in the sequence. In this experiment, we computed the variability of the following parameters: stride length and stride duration, using the same formula as Gabell and Nayak: CoVar = SD/mean x 100. The circumstances of the registration of the locomotion remained constant throughout the study. The same experimenter gave always the same instructions. All patients and controls were recorded in the same room, and at the same hour in the morning. Parkinson patients were fasting overnight, without treatment for at least 12 h. Statistical analysis Comparisons between Parkinson patients and controls were carried out using Mann-Whitney U-test. We used regression methods to test the relationships between velocity and locomotor parameters. The correlation between locomotor parameters and clinical stage were studied using Spearman rank correlation test (Siegel 1956).
RESULTS Compared to normal elderly controls, a slower walking velocity, a shorter stride length and a longer stride duration were observed in parkinsonian patients. Stance
94 duration and double support duration (absolute or relative to sequence duration) were longer in parkinsonian patients, whereas the swing phase was shorter; the value of the velocity peak was lower too. All these differences were significant. The average and the standard variations for each parameter and the z values with their significant thresholds for parkinsonian patients and controls are reported in Table 1. The relationship between the various parameters was preserved. In parkinsonian patients as well as in controls, there was a regression between the velocity of walking and stride length (R 2 = 0.91, y = 1.25x + 0.01; R 2 = 0.83, y = 0.86x + 0.26, respectively) (Fig. 1), relative duration of the double support (R 2 = 0.75, y = - 34x + 50; R 2 = 0.85, y = - 2 6 . 5 x + 43.8, respectively) (Fig. 2) and the peak velocity value (R 2 = 0.97, y = 3.27x + 0.55; R 2 -- 0.95, y = 3x + 0.80, respectively) (Fig. 3). C o m p a r e d to normal elderly controls (Table 1), the variability of stride duration was not modified in these parkinsonian patients (z = 0.122). Inversely, the variability of stride length was significantly higher in parkinsonian patients (z = 1.92, P < 0.05). There was a significant correlation between stride length variability and clinical stage (rs = 0.466; P < 0.05); stride length was more variable in type IV patients than in type I patients (U = 1.00; P < 0.05). There was also a significant correlation between the velocity of walking in parkinsonian patients and the H o e h n and Yahr stage (rs = 0.486; P < 0.05).
1.6 1.5 1.4 1.3 [31113'
1.2
0
o
1.1
[]
b~
+ 1
g
0.9 0.8 0,7 0,6 0.5 0.4 0.3 ~
0,20.1 0
1
0
I
i
0.2
I 0.4
I
I Q.6
I
[ Q.~
I
I
I
1
I 1 2 Velocity
+ ; Parkinsonians
I
I 1 4
1
J 1 5
(m/s)
~E3 : Controls
Fig. 1. By velocity representation of stride length in 21 parkinsonian patients ( + ) and 58 elderly controls (V]). Regression lines are y = 1.25x + 0.01; R 2 = 0.91 for Paxkinson patients and y = 0.86x + 0.26, R 2 = 0.83, for elderly controls.
95 60
50
+4-
40
20 o
D []
10
I
0
I
0.2
t
I
0.4-
I
I
0.6
I
I
I
I
0.8
I
1
I
[
1.2
[
1 .4
Velocity (m/s) + : Parklnsonians
Q : Controls
Fig. 2. By velocity representation of relative duration of the double support compared to the standard cycle in 21 parkinsonian patients ( + ) and 58 elderly controls (D). Regression lines are y = - 3 4 x + 50, R2 = 0.75 for Parkinson patients and y = - 26.5x + 43.8, R2 = 0.85 for elderly controls.
TABLE 1 MEAN AND SD FOR EACH PARAMETER AND THE z VALUES WITH THEIR SIGNIFICANT THRESHOLDS FOR 21 PARKINSONIAN PATIENTS AND 58 CONTROLS
Velocity (m/s) Stride length (m) Stride duration (sec) Swing duration (sec) Stance duration (sec) Swing velocity (m/s) Peak velocity Double support duration (sec) Relative double support duration (sec) Stride duration variability Stride length variability
PD patients (n = 21)
Controls (n = 58)
PD patients versus controls
Mean
SD
Mean
SD
z
P
0.44 0.57 1.29 0.40 0.88 1.36 2.00 0.23 35.8 5.51 6.68
0.20 0.26 0.16 0.08 0.16 0.44 0.66 0.08 9.80 5.10 4.27
0.83 0.97 1.19 0.46 0.74 2.12 3.28 0.13 21.8 4.75 5.17
0.23 0.22 0.15 0.05 0.12 0.46 0.72 0.05 6.80 2.30 3.80
5.25 4.95 2.17 2.88 3.50 5.03 5.31 5.21 5.24 0.12 1.92
< 0.01 < 0.01
1990, 98:91-97
91
Elsevier JNS 03357
Quantitative analysis of gait in Parkinson patients" increased variability of stride length O. Blin 1,2, A . M . Ferrandez 2 and G. Serratrice ~ 1Clinique des maladies du syst~me nerveux et de l'appareil locomoteur CHU Timone, Marseille (France), and 2URA CNRS 1166, Cognition et mouvement, tray. Susini, Marseille (France)
(Received 4 December, 1989) (Revised, received 19 March, 1990) (Accepted 20 March, 1990)
SUMMARY Analysis of the spatio-temporal and kinematic parameters of locomotion recorded in 21 parkinsonian patients compared to 58 normal elderly subjects showed significant differences in all the recorded parameters. However the relationship between these parameters was preserved, as was the basic locomotor pattern. The variability of stride length, more marked in parkinsonian patients, increased as a function of the clinical stages of Hoehn and Yahr. This index could be useful in assessing the course of the disease in patients.
Key words: Parkinson's disease; Locomotion; Quantitation
INTRODUCTION In spite of the frequency of gait disorders in Parkinson patients, few quantified studies of locomotion have been carded out. Differences between the locomotor parameters in parkinsonian patients and control subjects of the same age have already been reported (Knutsson 1972; Murray et al. 1978; Stem et al. 1983) for walking velocity, stride length and duration, stance duration and double support duration. Normally, there are strong correlations between velocity and different locomotor Correspondence to: Dr. O. Blin,Cliniquedes maladiesdu syst~menerveuxet de l'appareillocomoteur, CHU Timone, F-13385 MarseilleCedex5, France.
0022-510X/90/$03.50 © 1990Elsevier Science Publishers B.V. (BiomedicalDivision)
92 parameters such as stride length, relative double support duration (Ferrandez et al. 1988). These links may be called "invariant patterns". In fact, Stem et al. (1983) showed that "in parkinsonian patients the relative contributions of pace-length and cadence to the velocity of walking are no different from those in the elderly population at large". These authors even suggested quantifying only walking velocity to evaluate the locomotor disability of parkinsonians. Thus, certain invariants are preserved in parkinsonian patients. Locomotor parameters are different, but the relationship between the velocity and these parameters is normal. Besides, in these studies, no specific signs of Parkinson's disease have been detected in walking. Similar disorders have been observed in patients with very different pathologies (Imms and Edholm 1981). At the within-trial level, increased variability of movement exhibited by parkinsonian patients may be related to an increased variability in force production (Sheridan et al. 1987). Furthermore, Parkinson patients exhibit a force production impairment and have more irregular force-time curves than controls (Stelmach et al. 1989). The variability of some locomotor parameters was measured by Gabell and Nayak (1984) for a highly selected elderly population (only 32 out of 1187 subjects). They found no differences in variability for any of the parameters measured and concluded that any increases in variability of gait parameters must be due to pathology. We computed these variabilities on stride length and stride duration Parkinson patients. The relationship between progressive Parkinson's disease and the patient's walking difficulty has not been sufficiently investigated. Murray et al. (1978) showed that most locomotor parameters are related to the stages of the disease described in his clinical classification, although they admit that this classification is not very accurate. Stern et al. (1983) showed that a relationship exists between the different stages of the Hoehn and Yahr classification and factors such as walking velocity and duration of half turn. The aim of this study was: (1)to compare locomotor parameters and "invariant patterns" in parkinsonian patients and normal elderly controls; (2)to assess the intraindividual variability in parkinsonian patients and controls; (3)to compare parkinsonian patients to each other as a function of their clinical stage, and if possible also to determine evolutive and prognostic features.
MATERIALS AND METHODS
The study deals with 21 parkinsonian patients: 12 females and 9 males between the age of 50 and 85 years (average age 69.6 years), 1.50-1.85 m (average height 1.69 m), weighing between 37 and 88 kg (average weight 65 kg). The diagnosis of idiopathic Parkinson's disease was confn'med by a general and neurological examination. There were no neuromuscular, osteoarticular or psychiatric disorders in any of the subjects. The history of the disease ranged from 1 to 17 years and patients had been given L-DOPA treatment since the onset. According to the Hoehn and Yahr classification (1967) there were 4 patients in stage I, 3 patients in stage II, 7 patients in stage III, and 7 patients in stage IV. There were "freezing" episodes in 5 patients, dystonia in 3, dyskinesia in 5 and " O n - O f f ' episodes in 2.
93 Fifty-eight normal subjects, 23 males and 35 females between the age of 60 and 92 (average age 72 years), 1.52-1.82 m (average height 1.63 m), weighing between 40 and 120 kg without any neurological or rheumatologic disease or treatment liable to affect walking and balance, served as controls.
Procedure and recording The locomotor parameters were automatically recorded by means of an apparatus designed by Bessou et al. (1989). This device measures the longitudinal displacement of both feet during locomotion, and functions as follows. Each foot is attached to a separate string which is pulled as the subject walks. The string unwinds to the required length by means of a gearing-down pulley system. Movement is recorded in the form of an electric signal via a pulley linked to a potentiometer. The characteristics of locomotor displacement (stride length, cycle, stance, swing and double support durations, stride and swing velocities) are measured over a long distance (more than 10 m). This device does not require any particular walkway or specific illumination, and can be used to record normal locomotion. In fact the value of this device has been mainly demonstrated in pathological subjects by Bessou et al. (1989). In our study, date were recorded at 75 Hz, then filtered using a FIR filter with a 33-point window and a cutoff at 10 Hz. The calibration of the device defines the volt/meter coefficient to be used in computing spatial and velocity data. From these values, spatio-temporal data (stride length and stride duration, stance duration and double support duration) and kinematic data on locomotion were calculated (sequence velocity, swing velocity, peak velocity value). The variability index (as defined by Gabell and Nayak 1984) was used to assess intra-individual spread in the sequence. In this experiment, we computed the variability of the following parameters: stride length and stride duration, using the same formula as Gabell and Nayak: CoVar = SD/mean x 100. The circumstances of the registration of the locomotion remained constant throughout the study. The same experimenter gave always the same instructions. All patients and controls were recorded in the same room, and at the same hour in the morning. Parkinson patients were fasting overnight, without treatment for at least 12 h. Statistical analysis Comparisons between Parkinson patients and controls were carried out using Mann-Whitney U-test. We used regression methods to test the relationships between velocity and locomotor parameters. The correlation between locomotor parameters and clinical stage were studied using Spearman rank correlation test (Siegel 1956).
RESULTS Compared to normal elderly controls, a slower walking velocity, a shorter stride length and a longer stride duration were observed in parkinsonian patients. Stance
94 duration and double support duration (absolute or relative to sequence duration) were longer in parkinsonian patients, whereas the swing phase was shorter; the value of the velocity peak was lower too. All these differences were significant. The average and the standard variations for each parameter and the z values with their significant thresholds for parkinsonian patients and controls are reported in Table 1. The relationship between the various parameters was preserved. In parkinsonian patients as well as in controls, there was a regression between the velocity of walking and stride length (R 2 = 0.91, y = 1.25x + 0.01; R 2 = 0.83, y = 0.86x + 0.26, respectively) (Fig. 1), relative duration of the double support (R 2 = 0.75, y = - 34x + 50; R 2 = 0.85, y = - 2 6 . 5 x + 43.8, respectively) (Fig. 2) and the peak velocity value (R 2 = 0.97, y = 3.27x + 0.55; R 2 -- 0.95, y = 3x + 0.80, respectively) (Fig. 3). C o m p a r e d to normal elderly controls (Table 1), the variability of stride duration was not modified in these parkinsonian patients (z = 0.122). Inversely, the variability of stride length was significantly higher in parkinsonian patients (z = 1.92, P < 0.05). There was a significant correlation between stride length variability and clinical stage (rs = 0.466; P < 0.05); stride length was more variable in type IV patients than in type I patients (U = 1.00; P < 0.05). There was also a significant correlation between the velocity of walking in parkinsonian patients and the H o e h n and Yahr stage (rs = 0.486; P < 0.05).
1.6 1.5 1.4 1.3 [31113'
1.2
0
o
1.1
[]
b~
+ 1
g
0.9 0.8 0,7 0,6 0.5 0.4 0.3 ~
0,20.1 0
1
0
I
i
0.2
I 0.4
I
I Q.6
I
[ Q.~
I
I
I
1
I 1 2 Velocity
+ ; Parkinsonians
I
I 1 4
1
J 1 5
(m/s)
~E3 : Controls
Fig. 1. By velocity representation of stride length in 21 parkinsonian patients ( + ) and 58 elderly controls (V]). Regression lines are y = 1.25x + 0.01; R 2 = 0.91 for Paxkinson patients and y = 0.86x + 0.26, R 2 = 0.83, for elderly controls.
95 60
50
+4-
40
20 o
D []
10
I
0
I
0.2
t
I
0.4-
I
I
0.6
I
I
I
I
0.8
I
1
I
[
1.2
[
1 .4
Velocity (m/s) + : Parklnsonians
Q : Controls
Fig. 2. By velocity representation of relative duration of the double support compared to the standard cycle in 21 parkinsonian patients ( + ) and 58 elderly controls (D). Regression lines are y = - 3 4 x + 50, R2 = 0.75 for Parkinson patients and y = - 26.5x + 43.8, R2 = 0.85 for elderly controls.
TABLE 1 MEAN AND SD FOR EACH PARAMETER AND THE z VALUES WITH THEIR SIGNIFICANT THRESHOLDS FOR 21 PARKINSONIAN PATIENTS AND 58 CONTROLS
Velocity (m/s) Stride length (m) Stride duration (sec) Swing duration (sec) Stance duration (sec) Swing velocity (m/s) Peak velocity Double support duration (sec) Relative double support duration (sec) Stride duration variability Stride length variability
PD patients (n = 21)
Controls (n = 58)
PD patients versus controls
Mean
SD
Mean
SD
z
P
0.44 0.57 1.29 0.40 0.88 1.36 2.00 0.23 35.8 5.51 6.68
0.20 0.26 0.16 0.08 0.16 0.44 0.66 0.08 9.80 5.10 4.27
0.83 0.97 1.19 0.46 0.74 2.12 3.28 0.13 21.8 4.75 5.17
0.23 0.22 0.15 0.05 0.12 0.46 0.72 0.05 6.80 2.30 3.80
5.25 4.95 2.17 2.88 3.50 5.03 5.31 5.21 5.24 0.12 1.92
< 0.01 < 0.01
